chore: acts@44.4.0 +edm4hep

[wdconinc]

Briefly, what does this PR introduce?

This PR upgrades to acts@44.4.0 and ensures that we use acts +edm4hep which is what now contains the ActsPodioEdm podio data model definitions.

Note: the upgrade of acts to 44.4.0 is required because acts-project/acts#4820 fixes EDM4hepUtil.cpp to use edm4hep::CovMatrix6f brace initializers.

Needs:

• acts: patch for PR 4968, fix: Use (cached) runtime type hash in Acts::Any eic-spack#814

[veprbl]

Capybara is not happy: https://veprbl.github.io/capybara-reports/458fc303024e08f06439055c3579ce03/
there are changes to trajectories' nHoles, nOutliers as well as nStates. We probably should understand this better before we move forward with this.

[wdconinc]

Strange. acts-project/acts@v44.3.0...v44.4.0 looks harmless.

[veprbl]

The difference is introduced by acts-project/acts@6fb4b10

https://veprbl.github.io/capybara-reports/494fdd3dd92c2b93892ada4fdc7a0619/ is the comparisons between adjacent commits

[wdconinc]

The heck?!?

[wdconinc]

acts-project/acts#4858

[wdconinc]

acts-project/acts#4858

The ACTS issue actually fixes a bug that affected us (but we didn't know) for any TGeoTubeSeg placement of a tracking detector with a placement around 0 degrees (only MPGDInnerTracker). So it seems this is a change for the better.

We have some TGeoTubeSeg we place around 0 degrees, -dphi/2 to +dphi/2 style. These are given a phi = [-23.74,23.74] deg in the dd4hep Tube creation (i.e. we pass those values scaled in radians).

ROOT turns this into phi=[336.26,383.74] deg (in actual degrees), which previously (< v44.4) were then read by ACTS as [-0.41,6.70] old rad. With the fix this becomes (correctly) [-0.41,0.41] new rad.

Detailed logging output with the patch MPGDCylinderBarrelTracker_geo.patch against eic/epic@1f93f05

MPGDCylinderBarrelTracker INFO  Stave Model #0,"MMInnerSector": rmin = 55.50 cm  phi = [-23.74,23.74] deg
MPGDCylinderBarrelTracker INFO    Outward Frame Tube: phi=[336.26,383.74] deg = [-0.41,6.70] old rad = [-0.41,0.41] new rad
MPGDCylinderBarrelTracker INFO    Start/Stop -24.776553 -23.744197
MPGDCylinderBarrelTracker INFO    Start/Stop 23.744197 24.776553
MPGDCylinderBarrelTracker INFO  Stave Model #1,"MMOuterSector": rmin = 57.70 cm  phi = [-22.84,22.84] deg
MPGDCylinderBarrelTracker INFO    Outward Frame Tube: phi=[337.16,382.84] deg = [-0.40,6.68] old rad = [-0.40,0.40] new rad
MPGDCylinderBarrelTracker INFO    Start/Stop -23.831867 -22.838872
MPGDCylinderBarrelTracker INFO    Start/Stop 22.838872 23.831867
MPGDCylinderBarrelTracker INFO    Service to inner sector: rmin = 57.65 cm thickness = 0.05 cm length = 56.65 cm phi = [-21.55,21.55] deg

So, in summary, I think this is actually a change we should merge since it fixes an underlying issue.

[wdconinc]

Of course (since I buried the lede), the reason why this adds more holes to trajectories is because the cylinder surface bounds of the surface cover the entire phi circumference and then some. Every track will have a hole in that surface.

[wdconinc]

@mposik1983 FYI analysis of tracking issue affecting the MPGD inner barrel.

[wdconinc]

If we upgrade to ACTS v44.4.0, we will want to pick up acts-project/acts#4968 too (as a patch or as a potential v44.4.1) since it will affect our CI builds with gcc and clang.

[Copilot]

Pull request overview

Copilot reviewed 3 out of 3 changed files in this pull request and generated no new comments.

[veprbl]

As discussed in the tracking meeting and on Acts Mattermost, this is in fact fixing a bug.